A KEY factor in maintaining healthy life is balancing
energy intake and expenditure. Abnormalities in this
balance can lead to diseases, such as obesity, anorexia, and
other eating disorders, which may furthermore deteriorate into
chronic diseases if not seriously treated [1]. A crucial step to
solve the problems is to continuously measure daily calorie
balance [2]. There are many off-the-shelf solutions to measure
calorie expenditure, such as Fitbit, Philips DirectLife, etc.
However, continuously and non-invasively monitoring calorie
intake remains a challenge. Currently, the common solutions rely on users’ self-reports, which are neither convenient nor
precise since food intakes are versatile and energy contained in
different food may vary significantly [3]. It is highly desirable
to develop accurate and easy-to-use methods to monitor the
food eaten and predict energy intake. Although some food
intake monitoring methods exist, they are either inaccurate or
involve complex sensor systems, which limits their adoption
in daily life.